CLIF 2007
Columbus Large Image Format (CLIF) 2007 Dataset Overview
The CLIF dataset contains imagery from a large format EO platform. The data
was collected in October 2007. The scene in this dataset is a flyover of the
Ohio State University (OSU) Campus. The platform approaches the campus and
loiters over the area.
The sensor is a large format monochromatic electro-optical sensor comprised
of a matrix of six cameras. The matrix is 2 rows by 3 columns. Cameras 3, 1,
and 5 make-up the top row of the image, respectively. Cameras 2, 0, and 4
make-up the bottom row of the image. Each camera was oriented in such way as
to maximize coverage, yet allow enough overlap between images to help in
mosaiking the image to form a larger image. The cameras, for this data
collection, collect data at approximately 2 frames per second.
The CLIF Sample dataset is a single DVD containing:
- 300 (50 images x 6 cameras) raw files
- 50 auxiliary text files
- Matlab data viewer that reads the .raw and .txt file contents and
assembles the individual camera frames into a single (unregistered) image.
Data product above is available for immediate DOWNLOAD.
The full CLIF dataset (966 GB) contains:
- 96,702 (16117 images x 6 cameras) raw files
- 16117 auxiliary text files
- Frames from Professor James W. Davis (OSU) ground sensor network
- Matlab data viewer that reads the .raw and .txt file contents and
assembles the individual camera frames into a single (unregistered) image.
Challenge Problems:
The AFRL/Sensors Directorate is interested in novel research using this dataset, especially novel approaches to:
- MOSAICKING-Mosaicking (stitching) of the six cameras using both computer vision and photogrammetric approaches
- VIDEO REGISTRATION-Registering and stabilizing video
- GEOREGISTRATION-providing UTM coordinates for every pixel
- GIS FUSION-fusing the data with GIS information
- LAYERED REGISTRATION-Registering (Data Fusion Level 0) of the aerial and building sensor data
- TRACKING-Tracking vehicles
- ATR-Performing Automatic Target Recognition (ATR) on objects of interest